Total Energy Consumption Control based on Environmental ZSG- DEA

Total Energy Consumption Control based on Environmental ZSG- DEA Zuoren Sun School of Business, Shandong University (Weihai), Weihai 264209, Shandong, China Abstract This paper designs a new model to solve the problem of allocation for total energy consumption between provinces in China, based on the environmental production technology and ZSG-DEA method, comprehensively considering the input/output variables of energy, capital, labor, desirable and undesirable output. The results show that: Through the ZSG-DEA allocation of energy quotas, the efficiency in energy use will be promoted obviously; The average general regulation cost will reduce 94.472%, and average strict regulation cost will reduce 75.111%, but generate the control cost for energy consumption 116.946 billion RMB (deflated to 1952 price); ZSG-DEA allocation will reduce the environmental regulation cost, and generate cost for controlling energy consumption; The ZSG-DEA allocation will cause the equity losses. However, it will cause the slight influence on the carbon emission intensity, and make the energy quotas allocation compatible with the emissions reduction target. ey words: TOTAL ENERGY CONSUMPTION CONTROL, ENVIRONMENTAL ZSG- DEA, ENVIRONMENTAL REGULATION COST, COST OF TOTAL ENERGY CONSUMPTION CONTROL Introduction Total energy consumption control is the prerequisite for the feasibility of energy saving and emission reduction plan. The cost of energy saving and emission reduction can be effectively calculated only after total energy consumption is ascertained. Then the market-based trade of energy consumption quota can be possible and the minimum cost of total control be realized. However, at the present stage, central government s evaluation mechanism of energy saving for regional government takes energy consumption per unit of gross regional product (GRP) as a benchmark, which may result in problem of uncontrollable total energy consumption owning to uncertainty of GRP growth expectations. Therefore, models and methods need to be innovated to fulfill transition from intensity control to total control. In taking full account of the allocation of total energy consumption and energy consumption demand in each province, the problem of how to realize optimal output efficiency after energy consumption allocation needs further investigation in the meantime. Literature Review Studies on energy consumption quota have not been paid much attention both at home Metallurgical and Mining Industry, 2015, No. 5 113

and abroad for it is a relatively new issue. At present, only the target decomposition of renewable energy in European Union[1] and CO2 emission trading scheme supported to energy savings certificates[2] are concerned to this issue. By contrast,studies on allocation of carbon emission quota under total control have been intensively studied both at home and abroad. After the adoption of yoto Protocol, scholars from different countries have been advocating various methods on allocation of carbon emission quota, among which the convergence method, multisector convergence, multiple periods method by Netherland National Institute for Public Health and the Environment, alternative scheme based on reduction of carbon emission intensity by America, dual intensity target by orean scholars, sustainable development policy and measures by South Africa and gradual participation by Newell et al[3,4]. All these methods have provided inspiration on allocation of energy consumption quota. As Svendsen et al[5] suggested, total control is the principle method to achieve targets of pollutant emission reduction for each country while allocation of initial emission permits is the key issue involved. Specific to China s regional allocation of energy saving index, institutions like United States Department of Energy, Energy Department of National Development and Reform Commission (China), Development Research Center of the State Council (China) have done some work. However, most of the studies are focusing on direct analysis of energy intensity index. Investigation on energy quota under total energy control target is obviously insufficient. In conclusion, studies on regional allocation of energy-saving index have made great achievement. Relationship of energy-saving index between total energy consumption and energy intensity is further cleared. However, while it is recognized that regional allocation of energysaving index has a ZSG feature under total energy consumption constraint, there is little study on regional allocation of energy saving and emission reduction based on ZSG efficiency model. The employment of ZSG efficiency perspective will provide a quantitative basis for the scientific and economic implementation of the country s overall target on energy saving and emission reduction. In addition, the improvements on ZSG model will also enrich the methodology system of energy and environment policy analysis. ZSG-DEA Method based on Environmental Production Technology Input (or output) variable in classic DEA model (CCR,BCC etc. ) has relative high degree of freedom, while that adopted in resource allocation domain is often restricted to total sum as constant and constrained by ZSG. Increase (or decrease) of any DMU input (or output) variable corresponds to decrease (or increase) of other DMU input (or output) variable. Lins[6] first developed the ZSG-DEA model. Through redistribution of input (or output) variables by ZSG-DEA, efficiency value of every DMU is readjusted. All the DMU technology efficiency value reaches 1 and a new production frontier (ZSG-DEA production frontier) is formed. However, due to constraint of DEA efficiency model, traditional ZSG-DEA can only conduct efficiency calculation and quotas allocation on single input (or output). For example, in study initial allocation of CO2 emission right, Gomes[7] set CO2 as the only input variable, while population, energy consumption and GRP are set as output. Although this method is common in dealing with undesirable output, it violates the actual production process where energy consumption is taken as output variable and undesirable output as input variable. This paper attempts to improve the traditional ZSG-DEA method and sets CO2 as output variable (and add undesirable output constraint[8,9]) according to production theory while still sets energy consumption as input variable. It also aims to restructure the traditional single input/output ZSG-DEA Model (uni-input/output ZSG-DEA Model) based on non- parameter distance function and form environmental multiple input/output ZSG distance function model so to meet the requirement of single variable value distribution under multi inputs and outputs. Considering the constraints of total energy consumption control, this paper uses energy input oriented distance function: D(,,,, ) sup{ : ( /,,,, / e E LY C = λ E λ LY C λ) T} (1) Equation (1) denotes: given production technology (T), capital (), labor (L), GRP (Y), and CO2 (C), energy consumption should be reduced as much as possible. Equation (1) can be formulated as piecewise linear function: 114 Metallurgical and Mining Industry, 2015, No. 5

1 [ D( E,, LY,, C)] = min θ e st.. λ E θe, m= 1, 2,, M λ, n = 1, 2,, N λ L L, l = 1, 2,, L k km om k kn on k kl ol λ Y Y, i = 1, 2,, I k ki oi λ C = θc, j = 1, 2,, J k kj oj λ 0,, 2,, k (2) The variables in model (2) share the same definitions in model (1). It is different from traditional energy input oriented Shephard distance function lies in the addition of energy structure carbon intensity (CO2 emission per unit of energy consumption[10]) constraint. Based on model (2), total energy consumption constraint is added as well: D E LY C = λ E λ LY C λ T E λ = E (,,,, ) sup{ : ( /,,,, / ), / } e k k k k= 1 k= 1 (3) Equation (3) can be formulated as piecewise linear function: 1 D E LY C = θ [ (,,,, )] min e (4) Eom (1 θ ) st.. λk Ekm 1 + θeom, m= 1, 2,, M E λ, n = 1, 2,, N k kn on λ L L, l = 1, 2,, L k kl ol km k o λ Y Y, i = 1, 2,, I k ki oi λ C = θc, j = 1, 2,, J k kj oj λ 0,, 2,, k As shown above, Model (4) has characteristics of both environmental production technology and ZSG-DEA, and also can be use to solve the problem of optimal energy allocation under total consumption constraint. Assign the excessive energy input of province i to other provinces by using E (1 θ ) / E. At the same im i km k i time, because model (4) is nonlinear programming, the quantitative relation of technology efficiency θ i and θ i in model (4) and model (2) can be established through linear transformation: E (1 φθ) j ji i θ j W i = θi 1+ (5) E j j W In model (5), DMUs whose technology efficiency θ i is not 1 in model (2), make up cooperation set W. Equation φ = θ / ji j θi denotes proportional relation of technology efficiency between DMU j and DMU i. θ i expresses ZSG- DEA technology efficiency of each DMUs. Cost of Environmental Regulation and Total Energy Consumption Control The very important characteristic of environmental production technology is its concern to environmental regulation cost. Zhou et al[11] and Färe[12] studied the environmental regulation cost under weak disaposibility constraint of undesirable output. They assume that undesirable output is unavoidable in production process. The existance of undesirable output determins that desirable output is impossible to achieve the optimal value under weak disaposibility constraint. The gap between target value of undesirable output under weak disaposibility constraint and that without such contraint is cost of environmental regulation[13,14]. It can be formulated as model (6): D E LY C g g g E g LY g C g T (6) (,,,, ; E, y, u) = sup { β :( + β E,,, + β y, + β u) } where, g = ( ge, gy, g u) is direction vector. The different choice of direction vector may bring about four environmental regulations. General regulation is g = ( 0, Y,0). Strict regulation is g = ( 0, Y, C ). Adjustment of undesirable output only, is g = ( 0,0, C ). Direction vector without environmental regulation, undesirable output in assumption of strong disposability, is g = ( 0, Y,0). Under general regulation, the maximum potential desirable output of a certain DMU is at point F. Without environmental regulation, its desirable output is at point G. The distance between F and G is the cost of environmental regulation after addition of undesirable output under weak disposability constraint. Under strict regulation, the maximum potential desirable output of certain DMU is at point. The environmental regulation cost can be Metallurgical and Mining Industry, 2015, No. 5 115

represented by distance between point and I. As shown in figure 1, cost of I under strict regulation is higher than that of FG under general regulation. Figure 1. Cost of Environmental Regulation and Total Energy Consumption Control There is opportunity cost under total energy consumption control. That is, a DMU can choose to be out of the energy consumption quota allocation. It promises to meet efficiency requirements of ZSG-DEA (not involve itself in reference technology)by improving desirable output so to prevent itself from being deprived of energy consumption quota. This paper defines it as cost of total energy consumption control. In order to measure the target value of desirable output under alternative plan, Shephard output distance function is defined as follows: ' ' Dy ( E,, LY,, C) = inf { λ :( E,, LY, / λ, C) T '} (7) Where T' = { ( E,, LY,, C): ( Eo, o, Lo, Yo, Co) T } (8) The difference between production feasible set T' and T is that, T' is exclusive of target DMUs and it only aims to allocate ZSG-DEA energy consumption quota among other DMUs. Total energy consumption remains unchanged after allocation. Details equation (7) in form of DEA model is as follows: 1 ' [ D( E,, LY,, C)] = maxθ y st.. λθ E E, m= 1, 2,, M k= 1, k o k= 1, k o λ, n = 1, 2,, N λ L L, l = 1, 2,, L k kl ol k= 1, k o ' λk ki θ oi = k= 1, k o k= 1, k o k k km om k kn on Y Y, i 1, 2,, I λθ kc = C, j = 1, 2,, J k kj oj λk 0,, 2,, (9) In model (9), θ means technology efficiency of ZSG-DEA determined by model (4) which takes T' as reference technology. θe means energy consumption of every province after ZSG-DEA allocation of model (4). θc means CO2 emission of every DMU after allocation of model (2). θ Y means desirable output under ZSG-DEA constraints. (θ -1)Y means cost of total energy consumption control (distance between point M and F in figure 1). Data Source and Treatment 116 Metallurgical and Mining Industry, 2015, No. 5

We estimate total GDP in 2015 will be 63119.7 billion RMB (2010 constant price). The total energy consumption will be 4329.75 million tons of standard coal and the energy consumption per unit of GDP will be 0.686 tons of standard coal per 10,000 RMB (2010 constant price). In estimating GRP, CO2, energy consumption, capital stock and labor force input of every province in 2015, we make allocation among provinces according to proportion of each province s portion to national total on base year. National total CO2 emission is calculated according to 12th FYP emission reduction target. For sake of consistency with the quantity of value of capital stock, this paper transforms all the GRP-related data to 1952 constant price. On provincial CO2 emission calculation, it is estimated according to final energy consumption. Allocation of ZSG-DEA energy consumption quota By allocating each province s energy consumption quota based on model (4) and making comparison between traditional technology efficiency and ZSG-DEA technology efficiency, potential amount of energy saving before and after allocation can be calculated as shown in table 1: Table 1. Comparison between traditional DEA efficiency and ZSG-DEA allocation efficiency at provincial level in 2015 ZSG-DEA Traditional Traditional Potential Efficiency ZSG-DEA energy energy DEA energy after energy consumption consumption efficiency saving allocation saving Province quota (Ten thousand tons) (Ten thousand tons) (Ten thousand tons) (Ten thousand tons) Beijing 7963.359 1.000 0.000 10880.230 1.000-2916.871 Tianjin 7119.485 1.000 0.000 9727.256 1.000-2607.771 Hebei 30807.973 0.427 17644.652 17984.867 1.000 12823.105 Shanxi 18878.079 0.338 12504.146 8708.618 1.000 10169.461 Inner 18596.625 0.370 11714.211 9403.349 1.000 9193.276 Mongolia Liaoning 23163.930 1.000 0.000 31648.563 1.000-8484.634 Jilin 9329.821 0.586 3865.542 7465.771 1.000 1864.050 Heilongjiang 12685.807 0.637 4604.144 11041.868 1.000 1643.940 Shanghai 12565.426 1.000 0.000 17167.971 1.000-4602.545 Jiangsu 28736.020 0.930 2020.638 36500.865 1.000-7764.845 Zhejiang 18867.324 1.000 0.000 25778.169 1.000-6910.845 Anhui 10781.973 1.000 0.000 14731.264 1.000-3949.291 Fujian 10806.889 1.000 0.000 14765.306 1.000-3958.417 Jiangxi 7044.892 0.761 1683.818 7324.764 1.000-279.872 Shandong 39293.840 0.643 14010.403 34544.418 1.000 4749.423 Henan 23938.813 0.603 9511.467 19711.888 1.000 4226.926 Hubei 16614.080 1.000 0.000 22699.593 1.000-6085.513 Hunan 16157.436 0.722 4493.353 15936.479 1.000 220.957 Guangdong 29880.668 1.000 0.000 40825.552 1.000-10944.884 Guangxi 8574.968 0.777 1909.159 9107.404 1.000-532.436 Hainan 1493.834 1.000 0.000 2041.006 1.000-547.171 Chongqing 8520.252 0.681 2720.521 7924.094 1.000 596.158 Sichuan 19782.217 0.722 5495.992 19519.075 1.000 263.142 Guizhou 9170.512 0.326 6178.610 4087.795 1.000 5082.717 Metallurgical and Mining Industry, 2015, No. 5 117

Yunnan 9734.986 1.000 0.000 13300.780 1.000-3565.794 Shannxi 9748.971 0.694 2985.581 9240.728 1.000 508.243 Gansu 6643.785 0.371 4177.104 3370.194 1.000 3273.592 Qinghai 2846.024 0.338 1884.493 1313.726 1.000 1532.298 Ningxia 4105.891 0.263 3027.600 1473.254 1.000 2632.637 Xinjiang 9121.096 0.381 5644.422 4750.132 1.000 4370.963 After ZSG-DEA allocation, technology efficiency value of every province reaches 1. And Pareto optimality of energy allocation is realized under technology efficiency perspective. Eastern provinces like Guangdong, Liaoning, Jiangsu, Zhejiang, Hubei, Shanghai and Fujian, with higher traditional DEA efficiency, get more energy consumption quota after ZSG-DEA allocation. On the contrary, due to lower energy efficiency, most western provinces get less energy consumption after ZSG-DEA allocation comparing with original energy consumption. In particular, the total energy consumption of Shandong, Guizhou, Inner Mongolia, Shanxi and Hebei province would be decreased and accounts for 67% of the entire available transferable quotas. Consequently, the five provinces above are facing stronger energy constraint, which tally with the actual situation. Official data in China statistical yearbook, 2011 shows that provinces like Hebei, Inner Mongolia, Shanxi and Guizhou are at the end of the national energy efficiency ranking. Although Shandong province ranks eleven, in the middle of the ranking, its energy consumption is on the top. Therefore, it needs to reduce large amount of energy consumption in course of energy efficiency improvement. Original input and output combination of Jiangxi and Hunan province is closest to requirements of ZSG-DEA production frontier. As a result, the two provinces do not have much change after allocation of energy quota. This also explains that the input/output efficiency benchmark under ZSG-DEA allocation is in accordance with reference system of the two provinces traditional technology efficiency (Both provinces, located in central China, represent regions of moderate level of economic development. It is appropriate to take the two provinces as benchmark.). After ZSG-DEA allocation, provinces with higher efficiency can get extra energy consumption quota, while provinces with lower energy efficiency are deprived certain amount of energy consumption quota. On the model, it can be illustrated that ZSG-DEA production frontier shrinks further comparing to the original production frontier. Therefore, a new production frontier enfolded by original production frontier comes into being. Provinces are distributed on the newly formed ZSG-DEA production frontier and national optimal energy efficiency is realized. Table 2. Comparison on energy intensity, CO 2 emissions intensity and carbon intensity of energy structure before and after ZSG-DEA allocation at provincial level in 2015 Province E/GRP (before allocation) E/GRP (after allocation) CO 2 /GRP (before allocation) CO 2 /GRP (after allocation) CO 2 /E (before allocation) CO 2 /E (after allocation) (ton/10,0 00 RMB) (ton/10,00 0 RMB) (ton/10,00 0 RMB) (ton/10,00 0 RMB) (ton/ton ) (ton/ton ) Beijing 2.362 3.227 5.394 5.394 2.284 1.672 Tianjin 3.231 4.414 8.283 8.283 2.564 1.877 Hebei 6.323 3.691 15.802 6.752 2.499 1.829 Shanxi 8.589 3.962 18.441 6.226 2.147 1.572 Inner Mongolia 6.669 3.372 14.575 5.394 2.185 1.600 Liaoning 5.253 7.178 11.668 11.668 2.221 1.626 Jilin 4.506 3.606 11.144 6.527 2.473 1.810 Heilongjiang 5.121 4.458 9.963 6.347 1.945 1.424 118 Metallurgical and Mining Industry, 2015, No. 5

Shanghai 3.064 4.186 6.729 6.729 2.196 1.607 Jiangsu 2.904 3.688 7.070 6.573 2.435 1.782 Zhejiang 2.849 3.892 7.506 7.506 2.635 1.928 Anhui 3.652 4.989 9.463 9.463 2.591 1.897 Fujian 3.070 4.194 7.895 7.895 2.572 1.883 Jiangxi 3.120 3.244 7.088 5.394 2.272 1.663 Shandong 4.199 3.692 10.651 6.854 2.537 1.857 Henan 4.339 3.573 10.371 6.250 2.390 1.749 Hubei 4.355 5.951 11.440 11.440 2.627 1.922 Hunan 4.217 4.159 9.877 7.131 2.342 1.714 Guangdong 2.718 3.714 6.243 6.243 2.297 1.681 Guangxi 3.751 3.984 9.835 7.645 2.622 1.919 Hainan 3.029 4.138 6.537 6.537 2.158 1.580 Chongqing 4.500 4.185 10.141 6.903 2.254 1.649 Sichuan 4.818 4.754 9.292 6.711 1.928 1.411 Guizhou 8.341 3.718 19.217 6.270 2.304 1.686 Yunnan 5.641 7.707 14.150 14.150 2.509 1.836 Shaanxi 4.031 3.821 8.349 5.792 2.071 1.516 Gansu 6.749 3.423 16.209 6.018 2.402 1.758 Qinghai 8.822 4.072 17.880 6.041 2.027 1.483 Ningxia 10.172 3.650 23.343 6.130 2.295 1.680 Xinjiang 7.022 3.657 15.310 5.836 2.180 1.596 Average 4.914 4.210 11.329 7.203 2.332 1.707 Variance 4.259 1.082 20.031 4.046 0.042 0.022 Table 2 depicts energy utilization efficiency and CO 2 index changes after ZSG-DEA allocation. The mean value and variance of each province's energy intensity, CO 2 emission intensity and carbon intensity of energy structure after ZSG-DEA allocation is remarkably smaller than the value before allocation. The convergence effect of ZSG-DEA to CO 2 emission intensity index is particularly noticeable. This indicates, to a certain extent, that ZSG- DEA allocation has obvious target steering function in narrowing differences of energy intensity, CO 2 emission intensity and carbon intensity of energy structure. Moreover, energy saving after ZSG-DEA allocation and energy saving potential are consistent in distribution (see figure 2). It is strictly allocated according to each province s energy saving potential. 20000 Hebei 15000 Shanxi Shandong energy saving potential(10,000 tons) ZSG-DEA energy saving(10,000 tons) 10000 Inner Mongolia 5000 Jilin Henan Sichuan Xinjiang Hunan Gansu Jiangsu Jiangxi Guangxi Guizhou Ningxia Anhui Chongqing 0 Qinghai Beijing Zhejiang Fujian Shanghai Hainan Shaanxi -5000 Tianjin Yunnan Hubei -10000 Liaoning Guangdong -15000 Metallurgical and Mining Industry, 2015, No. 5 119

Figure 2. Potential energy saving and energy saving under ZSG-DEA constraint at provincial level in 2015 Comparing with traditional DEA, ZSG- DEA has constraint condition of total energy consumption. Double constraint of environmental regulation (mainly from weak disposability of undesirable output) and total Province energy consumption control (mainly from opportunity cost of energy consumption constraint) would be given. Based on formula from (6) to (9), this paper further considers cost of double constraint. Table 3. Cost of environmental regulation and total energy consumption control at provincial level in 2015 Total energy Environmental regulation cost Environmental regulation cost consumption before ZSG-DEA allocation after ZSG-DEA allocation control cost General environmental regulation cost Strict environmental regulation cost General environmental regulation cost Strict environmental regulation cost (100 million RMB) (100 million RMB) (100 million RMB) (100 million RMB) (100 million RMB) Beijing 0.000 0.000 0.000 0.000 0.000 Tianjin 0.000 345.220 0.000 345.220 1180.250 Hebei 6945.810 7627.430 0.000 156.670 1226.180 Shanxi 5455.880 5637.550 161.610 343.280 339.060 Inner 5085.430 5085.430 125.610 125.610 0.000 Mongolia Liaoning 269.560 3776.850 269.560 3776.850 5128.680 Jilin 1444.930 1682.910 0.000 46.230 434.670 Heilongjiang 2456.450 2693.070 507.050 743.670 437.700 Shanghai 204.820 767.900 204.820 767.900 1014.490 Jiangsu 108.160 1295.770 0.000 440.230 2162.270 Zhejiang 0.000 281.510 0.000 281.510 2592.730 Anhui 0.000 803.870 0.000 803.870 2227.200 Fujian 0.000 392.450 0.000 392.450 1632.250 Jiangxi 724.930 724.930 12.010 12.010 0.000 Shandong 4747.930 6164.520 0.000 232.500 2531.540 Henan 3743.330 4213.370 0.000 186.200 875.690 Hubei 0.000 1849.870 0.000 1849.870 4275.200 Hunan 1776.370 2478.500 0.000 576.010 1233.290 Guangdong 0.000 857.250 0.000 857.250 1729.920 Guangxi 390.280 949.730 0.000 141.390 954.160 Hainan 34.750 92.030 34.760 92.030 104.530 Chongqing 1184.450 1481.760 32.580 329.890 529.640 Sichuan 3268.300 3823.760 941.290 1496.750 1001.950 Guizhou 2604.910 2700.830 0.000 84.800 178.450 Yunnan 0.000 1622.790 0.000 1622.780 2801.410 Shaanxi 1530.760 1623.180 266.660 359.080 178.490 Gansu 1714.720 1774.750 0.000 6.150 113.840 Qinghai 843.720 864.150 45.830 66.260 38.670 Ningxia 1279.700 1309.000 0.000 27.110 55.090 Xinjiang 2456.560 2511.820 66.710 121.960 106.320 Average 1609.058 2181.073 88.950 542.851 1169.456 120 Metallurgical and Mining Industry, 2015, No. 5

Variance 48271.750 65432.200 2668.490 16285.530 35083.670 control of total energy consumption comes Table 3 shows cost of environmental into practice, it will have significant influence regulation is remarkably reduced after ZSGto its economic output. Comparing cost of DEA allocation. Provincial general total energy consumption control and environmental regulation cost is reduced from environmental regulation, total cost of total 4827.175 billion RMB before allocation to energy consumption control is higher than 266.849 billion RMB after allocation, while that of strict environmental regulation. This strict environmental regulation cost is reduced implies that total energy consumption control an outstandingly 75% from 6543.220 billion can reduce cost of environmental regulation RMB to 1628.553 billion RMB. Before effectively. Meanwhile, opportunity cost allocation, Hebei, Shandong, Shanxi and under energy consumption control will be Inner Mongolia has higher general and strict generated by ZSG-DEA allocation, which environmental regulation cost. Yet, the may influence the potential economic growth problem disappears after allocation. Liaoning of each province. province has higher cost of total energy By comparing energy intensity consumption control, which is related to the reduction target of each province in 12 characteristic of its industrial structure. th FYP Liaoning province is highly reliable to energy saving target decomposition scheme resources and petrochemical industry, which (draft), it shows that: make great contribution to its GRP. If strict Table 4. Comparison on ZSG-DEA energy consumption per unit of GRP and planning target in 12 th FYP at provincial level in 2015 Energy Energy intensity ZSG-DEA energy intensity in reduction in 12 th Planning Difference intensity in 2015 target 2010 FYP Province (ton/10,000 RMB 1952 constant price) (ton/10,000 RMB 1952 constant price) (%) (%) (%) Beijing 2.810 2.630 6.406 17.000-10.594 Tianjin 3.850 3.250 15.584 18.000-2.416 Hebei 7.530 4.680 37.849 17.000 20.849 Shanxi 10.220 4.380 57.143 16.000 41.143 Inner 7.940 7.620 4.030 15.000-10.970 Mongolia Liaoning 6.250 5.850 6.400 17.000-10.600 Jilin 5.360 5.480-2.239 16.000-18.239 Heilongjiang 6.100 4.870 20.164 16.000 4.164 Shanghai 3.650 3.080 15.616 18.000-2.384 Jiangsu 3.460 2.870 17.052 18.000-0.948 Zhejiang 3.390 2.860 15.634 18.000-2.366 Anhui 4.350 4.440-2.069 16.000-18.069 Fujian 3.650 3.740-2.466 16.000-18.466 Jiangxi 3.710 3.800-2.426 16.000-18.426 Shandong 5.000 4.680 6.400 17.000-10.600 Henan 5.170 3.850 25.532 16.000 9.532 Hubei 5.180 5.300-2.317 16.000-18.317 Hunan 5.020 4.100 18.327 16.000 2.327 Guangdong 3.240 2.730 15.741 18.000-2.259 Guangxi 4.470 4.290 4.027 15.000-10.973 Metallurgical and Mining Industry, 2015, No. 5 121

Hainan 3.610 3.500 3.047 10.000-6.953 Chongqing 5.360 4.260 20.522 16.000 4.522 Sichuan 5.740 4.980 13.240 16.000-2.760 Guizhou 9.930 4.880 50.856 15.000 35.856 Yunnan 6.720 6.440 4.167 15.000-10.833 Shannxi 4.800 4.420 7.917 16.000-8.083 Gansu 8.030 4.680 41.719 15.000 26.719 Qinghai 10.500 3.730 64.476 10.000 54.476 Ningxia 12.110 11.620 4.046 15.000-10.954 Xinjiang 8.360 8.110 2.990 10.000-7.010 Changes of energy intensity at provincial level under ZSG-DEA allocation are quite different from the 12 th FYP planning target. The reason is that the planning targets are determined by trend extrapolation according to each province's historical data. Along with adjustment of desirable economic growth, this may neglect the ZSG condition among provinces caused by fixed total energy consumption. Due to higher traditional DEA efficiency, the 11 provinces of Beijing, Tianjin, Liaoning, Shanghai, Zhejiang, Anhui, Fujian, Hubei, Guangdong, Hainan, and Yunnan can get more energy consumption quota after ZSG-DEA allocation. The reduction of their energy intensity is lower than the planned reduction target. On the other hand, provinces with lower traditional DEA efficiency like Hebei, Shanxi, Guizhou, Gansu and Qinghai are deprived of energy consumption quota. Therefore, these provinces confront to stronger energy consumption constraint and have sharp reduction of energy intensity. The reason is that ZSG-DEA allocation is based on totalfactor energy efficiency (comprehensively consider capital, labor, desirable and undesirable output). For convenience of analysis, humanities indicator (like total population, regional GDP per capita etc.) concerned with fairness is not involved. Tianjin, Shanghai, Jiangsu, Zhejiang and Guangdong are less sensitive to the control mode of total energy consumption. The gap between their energy intensity reduction and that of 12 th FYP target is about 2%. This is because the above mentioned five provinces have higher energy efficiency. According to technical efficiency, they get more energy consumption quota. In addition, they have higher energy consumption, quota allocation has less influence to them. In the Consultative draft, CO 2 reduction target falls into seven categories. Hereby, compare CO 2 emission intensity and reduction target in 12 th FYP under validity of ZSG-DEA technical efficiency: Table 5. Comparison on ZSG-DEA CO 2 emissions intensity and planning target in 12 th FYP at provincial level in 2015 Carbon ZSG-DEA carbon emission carbon emission Planning emission intensity intensity Difference intensity in 2010 in 2015 reduction in 12 th target Province FYP (ton/10,000 RMB 1952 constant price) (ton/10,000 RMB 1952 constant price) (%) (%) (%) Beijing 6.500 5.390 17.077 17.000 0.077 Tianjin 9.980 8.280 17.034 18.000-0.966 Hebei 19.040 10.490 44.905 17.000 27.905 Shanxi 22.220 7.730 65.212 16.000 49.212 Inner Mongolia 17.560 14.580 16.970 15.000 1.970 122 Metallurgical and Mining Industry, 2015, No. 5

Liaoning 14.060 11.670 16.999 17.000-0.001 Jilin 13.430 11.140 17.051 16.000 1.051 Heilongjiang 12.000 7.790 35.083 16.000 19.083 Shanghai 8.110 6.730 17.016 18.000-0.984 Jiangsu 8.520 6.960 18.310 18.000 0.310 Zhejiang 9.040 7.510 16.925 18.000-1.075 Anhui 11.400 9.460 17.018 16.000 1.018 Fujian 9.510 7.900 16.930 16.000 0.930 Jiangxi 8.540 7.090 16.979 16.000 0.979 Shandong 12.830 10.650 16.991 17.000-0.009 Henan 12.500 7.560 39.520 16.000 23.520 Hubei 13.780 11.440 16.981 16.000 0.981 Hunan 11.900 7.890 33.697 16.000 17.697 Guangdong 7.520 6.240 17.021 18.000-0.979 Guangxi 11.850 9.840 16.962 15.000 1.962 Hainan 7.880 6.540 17.005 10.000 7.005 Chongqing 12.220 7.890 35.434 16.000 19.434 Sichuan 11.200 7.890 29.554 16.000 13.554 Guizhou 23.150 9.840 57.495 15.000 42.495 Yunnan 17.050 7.710 54.780 15.000 39.780 Shannxi 10.060 7.520 25.249 16.000 9.249 Gansu 19.530 9.840 49.616 15.000 34.616 Qinghai 21.540 6.540 69.638 10.000 59.638 Ningxia 28.120 23.300 17.141 15.000 2.141 Xinjiang 18.450 15.310 17.019 10.000 7.019 Table 5 demonstrates that the difference between changes of carbon emission intensity at provincial level under ZSG-DEA allocation and 12 th FYP planning target is smaller than energy intensity difference. The reason lies in that, in researching process, this paper does not put constraint of CO2 emission. CO2 emission under ZSG-DEA allocation of each province depends on their traditional DEA efficiency value. Other provinces' influence to a certain province's CO 2 emission is out of consideration. However, under ZSG-DEA allocation, CO 2 emission intensity in Shanxi, Guizhou, Yunnan, Gansu And Qinghai province still decreases noticeably. In this paper, we considered the restriction of fixed parameter between energy and CO 2 emission. In adjustment process according to traditional energy efficiency, although allocation of energy consumption quota is not involved, sharp changes of energy consumption brings about sharp changes of CO 2 emission. Therefore, CO 2 emission intensity reduces dramatically. The gap between reduction target of CO 2 emission intensity and planning target in the 16 provinces of Beijing, Tianjin, Inner Mongolia, Liaoning and etc. is controlled within 2%. This implies that ZSG-DEA, in effective allocation of energy quota, has little interference to reduction target of CO 2 emission intensity. It can be policy compatible with implementation of planned target of CO 2 emission intensity. Although CO 2 emission intensity declines dramatically in Hebei, Shanxi, Guizhou, Gansu And Qinghai province, reduction target is not out of the reasonable range through comparison. Hebei's reduction target of CO 2 emission intensity in 2015 is about 10 ton/ 10,000 RMB, similar to Shaanxi's CO 2 emission intensity in 2010. Qinghai' s reduction target of CO 2 emission intensity in 2015 is about 7.8 ton/ 10,000 RMB, similar to Hainan's CO 2 emission intensity in 2010. Guizhou and Gansu's reduction target of CO 2 emission intensity in 2015 is about 9.8 ton/ 10,000 RMB, similar to Tianjin's CO 2 emission intensity in 2010. Shanxi's reduction target of CO 2 emission intensity is slightly higher, similar to Guangdong's CO 2 emission intensity in 2010. On the one hand, this demonstrates severe imbalance of energy input and output in Shanxi. Economic development pattern under "resource curse" has caused excessive reliance on energy input. On the other hand, the country should provide Shanxi with reasonable subsidy to improve total factor productivity and comprehensive output efficiency. Metallurgical and Mining Industry, 2015, No. 5 123

Conclusions This paper proposes a ZSG-DEA method based on environmental production technology, which having been improved by adopting nonparametric distance function, breaks single input or output restrictions of Gomes model. Through comprehensive consideration of allocation and substitution utility among factors of energy, capital and labor, efficient allocation of energy consumption quota under total factor perspective is achieved. Meanwhile, this paper pays more attention to influence of undesirable output to production efficiency and allocation plan. This enables the model to have a overall consideration to influence of environmental regulation during process of energy consumption quota. By applying ZSG-DEA model to study of "12 th FYP" total energy consumption control, its result demonstrates: (1) After ZSG-DEA allocation of total energy consumption, technology efficiency in every province is realized. Allocation of the amount of energy saving is also basically in accordance with each province's energy saving potential. (2) After allocation of ZSG-DEA energy consumption quota, environmental regulation cost is effectively reduced. Problem of total energy consumption control cost comes into being at the same time. (3) Allocation of energy consumption quota based on standard of ZSG- DEA technical efficiency is out of consideration on fairness. This explains why each province's total energy consumption control target after ZSG-DEA adjustment is not consistent with the planning target. However, constraint of ZSG-DEA total energy consumption control has little influence on planning reduction target of CO 2 emission intensity, which realizes the compatibility between energy saving and emission reduction policy to certain extant. In subsequent study, this paper will concentrate on how to realize consistency of total control target and energy intensity reduction target as well as the inter-temporal allocation based on time dimension. In modeling process, this paper lays stress on environmental DEA efficiency standard and takes energy, capital, labor input and effective relationship of desirable and undesirable output as standard for allocation of energy consumption quota. Due to the fact that provinces are heterogeneous individuals, who hold different features in industrial structure, economic development level, natural endowments, weather condition as well as energy utilization. It is not appropriate to simply compare each province's energy efficiency and calculate their energy saving potential from efficiency analysis perspective. In the future, this paper will make more effort to study the heterogeneous feature of each province's energy utilization in process of efficiency analysis. Acknowledgements This work was supported by the Natural Science Foundation of Shandong Province, China (Grant No. 2014ZRE27593), China Postdoctoral Science Foundation Grant (Grant No. 2014M561895) and Postdoctoral Science Foundation Grant of Shandong Province, China (Grant No. 201403015) References 1. Ramachandra, T.V. (2009) Riep: Regional Integrated Energy Plan. Renewable and Sustainable Energy Reviews, 13(2), p.p. 285-317. 2. Vine, E., Hamrin, J. (2008) Energy Savings Certificates: A Market-Based Tool for Reducing Greenhouse Gas Emissions. Energy Policy, 36(1), p.p. 467-476. 3. Newell, R.G., Stavins, R.N. (1999) Abatement Cost Heterogeneity and Anticipated Savings From Market-Based Environmental Policies. SG Working Paper No. 00-006, p.p. 4. Newell, R.G., Stavins, R.N. (2003) Cost Heterogeneity and the Potential Savings from Market-Based Policies. Journal of Regulatory Economics, 23(1), p.p. 43-59. 5. Svendsen, G.T., Vesterdal, M. (2003) How to Design Greenhouse Gas Trading in the Eu? Energy Policy, 31(14), p.p. 1531-1539. 6. Lins, M.P.E. (2003) Olympic Ranking Based On a Zero Sum Gains DEA Model. European Journal of Operational Research, 148(2), p.p. 312-322. 7. Gomes, E.G. (2007) Modelling Undesirable Outputs with Zero Sum Gains Data Envelopment Analysis Models. Journal of the Operational Research Society, 59(5), p.p. 616-623. 8. Zhou, P., Ang, B.W., Poh,.L. (2008a) Measuring Environmental Performance under Different Environmental DEA Technologies. Energy Economics, 30(1), p.p. 1-14. 9. Zhou, P., Ang, B.W., Poh,.L. (2008b) A Survey of Data Envelopment Analysis in Energy and Environmental Studies. European Journal of Operational 124 Metallurgical and Mining Industry, 2015, No. 5

Research, 189(1), p.p. 1-18. 10. Ang, B.W. (1999) Is the Energy Intensity a Less Useful Indicator than the Carbon Factor in the Study of Climate Change? Energy Policy, 27(15), p.p. 943-946. 11. Zhou, P., Ang, B.W. (2008) Linear Programming Models for Measuring -Wide Energy Efficiency Performance. Energy Policy, 36(8), p.p. 2911-2916. 12. Färe, R. (2007) Environmental Production Functions and Environmental Directional Distance Functions. Energy, 32(7), p.p. 1055-1066. 13. Chambers, R.G., Fāure, R., Grosskopf, S. (1996) Productivity Growth in APEC Countries. Pacific Economic Review, 1(3), p.p. 181-190. 14. Färe, R. (2005) Characteristics of a Polluting Technology: Theory and Practice. Journal of econometrics, 126(2), p.p. 469-492. Metallurgical and Mining Industry, 2015, No. 5 125